AMD Ryzen Branch Prediction Optimizations Now Available to Windows 11 23H2

[cerulliber]

Gaming community has already embraced Win11 by looking at Steam numbers (as flawed as they are being opt-in).

Steam Survey July 2024 Update: Windows 10 Usage Records Uptick, Windows 11 Drops | TechPowerUp
TPU wroted that Windows 10's share rose to 47.69%, marking a significant uptick that contrasts with Windows 11's decline to 45.73%
once they push KB5041587 as mandatory update or release 24H4 home/pro rtm situation will change after win11 is faster youtube videos.
KB5041587 it's an optional update, so they not pushing it yet to public.
HUB benchmarks arent' by updating stable windows, they benchmarked by downloaded beta 24H2 which is not stable yet. moreover, for 24h2 iot enterprise rtm(factory debloated) there is no KB5041587 . this one is the only 24H4 I've found not beta (RTM).
if you want to replicate HUB methology, first step is to install an beta windows.which I don't want to.
also, please note that in hardwareluxx benchmarks, 5800x3d benefits too.

 

[Tomorrow]

Steam Survey July 2024 Update: Windows 10 Usage Records Uptick, Windows 11 Drops | TechPowerUp
TPU wroted that Windows 10's share rose to 47.69%, marking a significant uptick that contrasts with Windows 11's decline to 45.73%
once they push KB5041587 as mandatory update or release 24H4 home/pro rtm situation will change after win11 is faster youtube videos.
KB5041587 it's an optional update, so they not pushing it yet to public.
HUB benchmarks arent' by updating stable windows, they benchmarked by downloaded beta 24H2 which is not stable yet. moreover, for 24h2 iot enterprise rtm(factory debloated) there is no KB5041587 . this one is the only 24H4 I've found not beta (RTM).
if you want to replicate HUB methology, first step is to install an beta windows.which I don't want to.
also, please note that in hardwareluxx benchmarks, 5800x3d benefits too.

Windows share fluctuates several percent month-to-month in Steam HW survey.
I dont put too much faith into these one month changes.

 

[Lew Zealand]

0.01% difference in Horizon Zero Dawn, no difference in CPU or Average FPS.
I ran the bench 3 times each, these were the best of each

You're heavily GPU-limited. Your CPU can push 3x those frames with ease in H:ZD as it's an older console-based game. Which is a good point though as most of us are GPU-limited so we'll only see the CPU benefits in those rare CPU-heavy or CPU-broken games. And then maybe only in the 1% lows, which is still a good thing.

 

[billeman]

Also not offering 23H2 seems to be happening on systems where users have either disabled TPM and/or circumvented other requirements to install Win11. At least that's my impression so far, as my own system was one of those (it's perfectly capable of supporting TPM etc but i chose to disable it).

In my case both systems are fully win11 compliant, I guess it's because I use a local windows account to login and not a Microsoft account.

 

[Tomorrow]

In my case both systems are fully win11 compliant, I guess it's because I use a local windows account to login and not a Microsoft account.

Good point. I too use a local account. The reason i suspected system requirements was because MS at one point said that systems that are installed by bypassing these requirements may not receive updates, but they never clarified any further.

 

[529th]

W11 22H2, OS build 22621.4112

W11 23H2, OS build 22631.4112

Both with KB5041587

 

[Tomorrow]

Does anyone know if KB5041587 offers the same benefits on 22H2 as 23H2? I got that small update but failed to get baseline bench results before updating for comparison.

You could use System Restore, unless it's disabled or it is enabled but you have no previous restore points available.

 

[Shadowsnight]

The game says that you are using win 10 pro though.

Yeah it just says that, the HWinfo screenshot shows the winver I'm using XD

You're heavily GPU-limited. Your CPU can push 3x those frames with ease in H:ZD as it's an older console-based game. Which is a good point though as most of us are GPU-limited so we'll only see the CPU benefits in those rare CPU-heavy or CPU-broken games. And then maybe only in the 1% lows, which is still a good thing.

I just ran the test with the lowest possible settings, vsync off (it says custom because I turned off anti-aliasing) CPU has been consistently the slowest stat in this benchmark. I think I just assumed it was always the bottleneck, I guess I should have run it like this before the update my bad XD

 

[Imouto]

*snip*

You really speak like a console user trying to play down PC gaming saying it's too complicated.

 

[Lew Zealand]

Yeah it just says that, the HWinfo screenshot shows the winver I'm using XD

I just ran the test with the lowest possible settings, vsync off (it says custom because I turned off anti-aliasing) CPU has been consistently the slowest stat in this benchmark. I think I just assumed it was always the bottleneck, I guess I should have run it like this before the update my bad XD

Woo 264 FPS!

H:ZD's benchmark does not identify the CPU's and GPU's individual performance properly and I don't understand why, as Shadow of the Tomb Raider and Forza Horizon 4 and others can do this. Especially as it still makes individual fps claims for each. They just should not be there at all.

Nah no bad for you, I've just run this benchmark on too many systems at this point (dozens of different setups) and now know the individual CPU and GPU fps numbers are simply broken.

 

[AtarSt]

ugh this wording again. Thats not how branch prediction works. This reads like all the other outlets in that the "Administrator" account is doing something to the branch predictor, it is NOT the % improvements are just similarly comparable.

I encourage anyone to look up what branch prediction is before thinking it has anything to do with being an "Admin".

this stuff makes these threads super active so I get it but damn.

Couldn't it be related because an admin running code would not have to be as careful to avoid branch related exploits (meltdown spectre etc.), so the branch prediction/next instruction pipelining could be done more aggressively on admin vs a normal user? I don't know how it actually works in windows but this seems to me like it could be a possibility.

 

[Chrispy_]

You really speak like a console user trying to play down PC gaming saying it's too complicated.

I work in IT. I spend all day every day explaining stuff to the masses. Trust me when I say that Linux is scary for the overwhelming majority of the population even if you or I have no issues with it. When I say what I say, it's not my opinion, it's real-world experience of thousands of staff who have come to me for help over the last 25 years in the industry.

 

[sixor]

Amd fine wine for cpus

Free performance, thanks, i will take it

Just as the awesome dx12 perf for amd 580s with async compute

 

[Laurijan]

Amd fine wine for cpus

Free performance, thanks, i will take it

Just as the awesome dx12 perf for amd 580s with async compute

I take extra AMD performance anytime over Spectre / Meltdown update Intel robbing my of 10% of my CPU performance

 

[Chrispy_]

I take extra AMD performance anytime over Spectre / Meltdown update Intel robbing my of 10% of my CPU performance

Well, that free performance might get undone with mitigation patches for Sinkclose at some point.

It's not something you need to consider, just like you don't have to worry about Spectre/Meltdown in processors less than 5 years old today. You have far bigger problems to worry about with Intel these days than Spectre/Meltdown mitigations, anyway!

 

[Wirko]

You have far bigger problems to worry about with Intel these days than Spectre/Meltdown mitigations, anyway!

That's true even if you are Intel!

 

[Marcus L]

Installed latest version of Windows 11 Pro (23/H2?) to get this download, completely borked my start menu, couldn't type in start menu would start typing and it would only recognise the 1st letter I typed and send me to the 1st app in start menu that began with that letter, 2nd letter wasn't recognised at all, was completely broken, tried a few CMD/PowerShell commands to reset the start menu and nothing worked for me, had to do a roll back of the OS, thinking of going back to 10, sick of this beta shit

 

[lexluthermiester]

The game says that you are using win 10 pro though.

That doesn't mean anything because...

Could be misreading Win11 as Win10.

...a lot of programs are doing this.

 

[RJARRRPCGP]

That doesn't mean anything because...

...a lot of programs are doing this.

And Unigine Superposition shows Windows 10, because it's from the late-2010s. Way before ray tracing became mainstream. Seemed like only someone could dream about ray tracing in their sleep at that time!

 

[Melvis]

Just got a Update for Windows 10 22H2, is that what this update is for Windows 11? but for 10? Just wondered!

 

[529th]

Just got a Update for Windows 10 22H2, is that what this update is for Windows 11? but for 10? Just wondered!

The update I think you're looking for is KB5041587

These are not the droids you're looking for.

 

[RJARRRPCGP]

Just got a Update for Windows 10 22H2, is that what this update is for Windows 11? but for 10? Just wondered!

I'm coincidentally running Windows 10 22H2 on my X58 system right now, to check an HDD that's old enough that I think there's a severe incompatibility issue with late SATA chipsets. That's why I'm using a SATA 2 system to scan with Victoria. The SATA port keeps on resetting, according to Windows 11 on the A320 chipset system in the other room, the HDD SMART SATA CRC error rate keeps climbing, but that doesn't happen on the 1366.

Looks like the HDD is from 2006. That can explain why it don't want to talk with SATA 3 SATA controllers!

 

[L'Eliminateur]

I'm coincidentally running Windows 10 22H2 on my X58 system right now, to check an HDD that's old enough that I think there's a severe incompatibility issue with late SATA chipsets. That's why I'm using a SATA 2 system to scan with Victoria. The SATA port keeps on resetting, according to Windows 11 on the A320 chipset system in the other room, the HDD SMART SATA CRC error rate keeps climbing, but that doesn't happen on the 1366.

Looks like the HDD is from 2006. That can explain why it don't want to talk with SATA 3 SATA controllers!

that's... not how SATA works, sata is completely backwards compatible, you can connect the first ever SATA HDD to a modern sata 3 controller and it will have absolutely no problem as it will link at SATA 1 speeds.
You either have a cabling problem or a hardware issue on that port, it could be that you came on some edge case incompatibility but it's odd, ¿did you disable SATA spread spectrum?, shitty setting that always did more problems than good.

I have older HDDs that i have connected to my X670 with no issue

 

[RJARRRPCGP]

that's... not how SATA works, sata is completely backwards compatible, you can connect the first ever SATA HDD to a modern sata 3 controller and it will have absolutely no problem as it will link at SATA 1 speeds.
You either have a cabling problem or a hardware issue on that port, it could be that you came on some edge case incompatibility but it's odd, ¿did you disable SATA spread spectrum?, shitty setting that always did more problems than good.

I have older HDDs that i have connected to my X670 with no issue

I didn't look for the SATA spread spectrum setting, but the same SATA cable is OK with other drives! I also unplugged the SATA cable and reinserted it, like I have to with USB drives countless times!

The suspect HDDs, are before 2009-ish, maybe <2008?

 

[L'Eliminateur]

my X670 does not have SATA spread spectrum anymore iirc, i'd still try other sata cables, they can be kooky sometimes .
Check if the drives have some weird jumper setting, early SATA drives still had jumpers, sometimes for spread spectrum others for power up in standby, remove them all or leave it as default